Crane non-probability reliability collaborative evaluation method based on multi-source data fusion

Abstract

The invention discloses a crane non-probability reliability collaborative evaluation method based on multi-source data fusion, which ingeniously collects multi-source data when a crane works, maps the collected data into intervals and performs weighted fusion, breaks through the limitation of single parameter evaluation, and improves the reliability of the crane. The operation state of the crane can be comprehensively reflected in multiple dimensions, and the comprehensiveness and credibility of an evaluation result are improved; on the basis, according to the scheme, the consistency among the features is analyzed and quantified through interval correlation, then the weight is generated through the average correlation degree, discount processing is carried out on high-conflict data, interference of abnormal or noise information on final evaluation is effectively restrained, and the robustness of the evaluation method under the complex working condition is enhanced; according to the scheme, interval information is converted into mass distribution by adopting a non-probability evidence theory, dependence on a large number of historical fault samples is avoided, and the method is suitable for crane safety evaluation with limited data or variable working conditions.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of crane working condition evaluation and crane safety management, and particularly relates to a crane non-probabilistic reliability collaborative evaluation method based on multi-source data fusion. BACKGROUND

[0002] As a key large-scale equipment in the industries of port, metallurgy, building material and machinery manufacturing, the safe operation of a crane is directly related to production efficiency and personnel and property safety. Traditional crane state monitoring and safety evaluation mostly rely on a single type of sensor or parameter, such as stress curve, vibration spectrum or temperature change for evaluation. Typical methods include mechanical model evaluation based on limit state design, fault diagnosis based on vibration signals, and thermal safety monitoring based on temperature threshold, etc.

[0003] In general, the following shortcomings exist in the evaluation of the working safety condition of a crane by using a single parameter: (1) information island, easily affected by single point failure; (2) only relying on stress signals obtained by a single sensor (such as a strain gauge), unable to reflect other working condition information such as vibration or temperature, once the sensor fails or the data is abnormal, the evaluation result deviates greatly from the real working condition; (3) strong model dependence, requiring a large amount of historical data, such as reliability prediction based on probability statistics or machine learning, requiring a large number of fault and normal operation samples for training, with high data acquisition cost and difficult to meet in new equipment or complex working conditions; (4) insufficient conflict handling, poor robustness, when multiple parameters are evaluated, a weighted average or confidence superposition is often simply used, it is difficult to effectively handle the conflicts or inconsistencies between different sensor data, and there is no self-adaptive mechanism for weight determination, which is easy to cause the evaluation result to be sensitive to individual high conflict data.

[0004] In recent years, "non-probabilistic reliability" and "Dempster-Shafer Theory" (D-S) have attracted attention in the field of engineering reliability evaluation. The non-probabilistic method describes uncertainty through interval, fuzzy, etc. without prior assumption of random distribution; while the D-S evidence theory can fuse evidences with different trust degrees from various sources, and when there is conflict, it can discount or redistribute high conflict evidences through conflict factor. However, existing applications are mostly limited to a single non-probabilistic model or do not combine multi-source data conflict degree for weight optimization, and there is still no non-probabilistic collaborative evaluation system for cranes that can dynamically update in real time under multi-sensor, multi-working condition conditions. Therefore, there is an urgent need for a scheme that can simultaneously fuse multi-type sensor data such as stress, vibration and temperature, to break through the dependence on a single parameter or a single model in the prior art, and to improve the comprehensiveness and reliability of crane safety evaluation under complex working conditions. SUMMARY

[0005] Therefore, the present application aims to provide a crane non-probabilistic reliability collaborative evaluation method based on multi-source data fusion, which can integrate multiple types of sensor data, has good robustness and safety state evaluation reliability.

[0006] To achieve the above technical purposes, the technical scheme adopted by the present application is:

[0007] A crane non-probabilistic reliability collaborative evaluation method based on multi-source data fusion comprises:

[0008] S01, when the crane is in a working state, collecting monitoring data fed back by a sensor group on the crane, and then preprocessing the monitoring data;

[0009] S02, within a preset time window, defining a data interval representation of each type of sensor respectively to determine the upper and lower limits of the data within the corresponding time interval and obtain a fusion interval;

[0010] S03, for any two fusion intervals, calculating the interval correlation degree thereof, and also calculating the average correlation degree of each fusion interval with other intervals;

[0011] S04, mapping each fusion interval to a non-probabilistic quality assignment for a safety state to complete non-probabilistic modeling;

[0012] S05, calculating a weighting coefficient according to the average correlation degree of each fusion interval, and weighting the quality assignment;

[0013] S06, sequentially fusing all weighted quality assignments according to the D-S evidence theory to eliminate conflicts and obtain a final synthetic quality assignment function;

[0014] S07, calculating the confidence and the confidence degree according to the final synthetic quality assignment function and the non-probabilistic modeling, and then calculating a comprehensive reliability index;

[0015] S08, judging the working safety condition of the crane according to the comprehensive reliability index.

[0016] As a possible implementation manner, further, in the present application S01, the monitoring data comprises one or more of stress, vibration and temperature, which are collected and fed back by different types of sensors respectively.

[0017] As a possible implementation manner, further, in the present application S02, when the same monitoring data is collected by multiple same type sensors, the data collected and fed back by the same type sensors within the same preset time window are weighted and fused according to a preset weight to obtain a fusion interval.

[0018] wherein the data interval representation of each type of sensor is is the sensor number, are respectively the lower limit value and the upper limit value of the data collected by sensor j within the preset time window;

[0019] When the same monitoring data is collected by a single sensor, the fusion interval I r is expressed as follows: i ;

[0020] When the same monitoring data is collected by multiple sensors of the same type, the fusion interval I r is expressed as follows:

[0021]

[0022] where j is the sensor number, are respectively the lower limit value and the upper limit value of the data collected by sensor j within the preset time window; w j is the weight value of the data collected by sensor j, and the cumulative result of the weight values corresponding to multiple sensors of the same type is 1, i.e., N is the number of multiple sensors of the same type.

[0023] As a preferred implementation option, preferably, in the scheme S01, the pre-processing manner of the monitoring data includes using an SG filtering algorithm to perform filtering and smoothing processing on the monitoring data, and performing normalization processing on the data after filtering and smoothing processing.

[0024] where the monitoring data collected by the sensor in feedback is recorded in sequence as X = {x i |i = 1, 2, …, m}, x i is the monitoring data collected by the sensor in feedback at the i-th moment, and m is the number of data samples; the data sequence after filtering and smoothing processing is represented as:

[0025]

[0026] where M ≤ i-1 ≤ m-M, the filtering window width is 2M+1, c k is the SG filtering coefficient, and satisfies ∑c k = 1, is the smoothed value of the monitoring data at the i-th moment after filtering; the data record obtained after filtering and smoothing processing is

[0027] The normalization processing on the data after filtering and smoothing processing includes:

[0028] The data is subjected to minimum-maximum normalization processing within the same time window, and the formula definition is as follows:

[0029]

[0030] where is the filtered and smoothed, normalized monitoring data value at the i th moment, is the filtered and smoothed monitoring data value at the i th moment,

[0031] are the minimum and maximum values of the smoothed data sequence

[0032]

[0033] As a preferred implementation option, preferably, the present scheme S03 comprises:

[0034] Any two fusion intervals in the fusion interval are set as

[0035] The interval correlation degree ε of each fusion interval is calculated ab , which is defined as follows:

[0036]

[0037] Wherein, ε ab ∈ [0, 1], the greater the value of ε ab , the more overlap between the fusion intervals I a and I b , the stronger the consistency between them.

[0038] On the basis of the above, the average correlation degree of each fusion interval with other intervals is also calculated which is defined as follows:

[0039]

[0040] Wherein, K is the total number of monitoring data types to be analyzed after data fusion; is the consistency average of the k th fusion interval with other fusion intervals.

[0041] As a preferred implementation option, preferably, the present scheme S04 comprises:

[0042] Each fusion interval is set as is mapped to the non-probabilistic mass assignment m of the safety state H k to complete non-probabilistic modeling, which is defined as follows:

[0043]

[0044] Wherein, S max is the preset safety upper limit value, F max is the preset failure threshold, are the lower and upper limits of the reliability mapped by the fusion interval, respectively; m​k (H) the quality of the evidence supporting “safety” for category k, is the quality of the k-th type of data evidence supporting “failure”, m k (θ) is the mass representing uncertainty.

[0045] As a preferred implementation option, preferably, this solution S05 includes:

[0046] According to the average correlation of each fusion interval Calculate the weighting coefficient α k , which is defined as follows:

[0047]

[0048] Among them, α k is the weight factor of the kth piece of data evidence, that is, the weight factor of the monitoring data collected by the kth sensor after fusion, which is used to assign quality m k Weighted, α k ∈[0,1], and satisfies K is the total number of monitoring data types to be analyzed after data fusion; is the average consistency between the k-th fusion interval and other fusion intervals, j is an index variable, which represents the traversal of K data evidences for calculating the normalized denominator of the weight factor, In order to normalize the inconsistency of all data evidence,

[0049] In this scheme, the weighting coefficient α k Assign mass m k The weighting is defined as follows:

[0050] m′ k (A) = α k m k (A)

[0051] in, m′ k (A) is the weighted mass distribution, α k is the weight factor of the kth piece of data evidence, m k (A) is the quality distribution of the k-th piece of data evidence.

[0052] As a preferred implementation option, preferably, this solution S06 includes:

[0053] The conflict factor of any two pieces of data evidence is defined as follows:

[0054]

[0055] in, K ij is a conflict factor, m′ i (B), m′ j (C) is the weighted quality assignment of the ith and jth data evidence, respectively.

[0056] In the scheme, all weighted quality assignments are fused in turn according to the D-S evidence theory to eliminate conflicts, wherein the function for fusing the quality assignments of two data evidences is defined as follows:

[0057]

[0058] wherein, m ij (A) is the quality assignment after re-fusing the weighted quality assignments of the ith and jth data evidences.

[0059] In the scheme, the fusion of the quality assignments of data evidences is extended to K data evidences, and two-by-two merging is performed in turn to finally obtain the final synthetic quality assignment function m, which is defined as follows:

[0060]

[0061] wherein, m′ K is the weighted quality assignment of the Kth data evidence.

[0062] As a preferred implementation option, the scheme S07 preferably comprises:

[0063] According to the final synthetic quality assignment function m and the non-probabilistic modeling, the belief Bel(H) and the plausibility PI(H) are calculated, which are defined as follows:

[0064]

[0065] According to the calculated belief and plausibility, the comprehensive reliability index is further calculated, which is defined as follows:

[0066]

[0067] wherein, Bel(H) ∈ [0, 1], which represents the safety confidence in the most conservative state, PI(H) ∈ [0, 1], which represents the safety confidence in the most optimistic state, and the comprehensive reliability index R ∈ [0, 1].

[0068] As a preferred implementation option, the scheme S08 preferably comprises:

[0069] The corresponding relationship between the comprehensive reliability index and the working safety state of the crane is defined, which includes the following:

[0070]

[0071] Judge the working safety condition of the crane according to the comprehensive reliability.

[0072] Based on the above, the scheme also proposes a crane non-probabilistic reliability collaborative evaluation system based on multi-source data fusion, which includes:

[0073] The data collection module collects the monitoring data fed back by the sensor group on the crane when the crane is in the working state, and then pre-processes it;

[0074] The data fusion module is used to define the data interval representation of each type of sensor within a preset time window, to determine the upper and lower limits of the data within the corresponding time interval, and to obtain the fusion interval;

[0075] The data processing unit is used to calculate the interval correlation of any two fusion intervals, and also calculates the average correlation of each fusion interval with other intervals;

[0076] The data modeling module is used to map each fusion interval to a non-probabilistic quality assignment for the safety state, to complete non-probabilistic modeling;

[0077] The data post-processing module calculates the weighting coefficient according to the average correlation of each fusion interval, and weights the quality assignment;

[0078] The data synthesis module is used to sequentially fuse all weighted quality assignments according to the D-S evidence theory, to eliminate conflicts, and to obtain the final synthesized quality assignment function;

[0079] The reliability calculation module is used to calculate the belief and plausibility according to the final synthesized quality assignment function and non-probabilistic modeling, and then to calculate the comprehensive reliability index;

[0080] The safety evaluation module is used to judge the working safety condition of the crane according to the comprehensive reliability index.

[0081] Compared with the prior art, the application has the beneficial effects that: the scheme ingeniously collects multi-source data (stress, vibration, temperature) and the like of the crane in operation, maps the collected data into intervals and performs weighted fusion, breaks through the limitation of single parameter evaluation, can comprehensively reflect the running state of the crane in multiple dimensions, and improves the comprehensiveness and reliability of the evaluation result; on this basis, the scheme further quantifies the consistency between the features by interval correlation analysis, generates weights by "average correlation degree", discounts high conflict data, effectively suppresses the interference of abnormal or noise information on the final evaluation, and enhances the robustness of the evaluation method under complex working conditions; the scheme converts interval information into quality distribution by using non-probabilistic evidence theory, does not depend on a large number of historical fault samples or pre-distribution assumptions, is suitable for crane safety evaluation with limited data or variable working conditions, provides a feasible scheme for crane evaluation in new equipment or first working condition, in addition, the scheme combines weighted discount and pairwise synthesis strategy on the basis of the traditional Dempster-Shafer rule, effectively manages evidence conflict, and ensures the rationality and stability of the fusion result; at the same time, the fusion process explicitly calculates the conflict factor, and enhances the interpretability of the method. BRIEF DESCRIPTION OF DRAWINGS

[0082] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only constitute some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without any creative effort.

[0083] Figure 1 is a brief implementation process schematic diagram of the crane non-probabilistic reliability collaborative evaluation method based on multi-source data fusion of the present scheme;

[0084] Figure 2 is a unit module connection diagram of the crane non-probabilistic reliability collaborative evaluation system based on multi-source data fusion of the present scheme. DETAILED DESCRIPTION

[0085] The present application will be further described in detail below in combination with the drawings and embodiments. It is particularly pointed out that the following embodiments are only used to illustrate the present application, but do not limit the scope of the present application. Similarly, the following embodiments are only part of the embodiments of the present application, not all embodiments, and all other embodiments obtained by those skilled in the art without any creative effort are within the scope of the present application.

[0086] As Figure 1As shown, this embodiment provides a crane non-probabilistic reliability collaborative evaluation method based on multi-source data fusion, which includes:

[0087] S01. When the crane is in operation, collect monitoring data fed back by the sensor group on the crane and then pre-process it;

[0088] S02. Within a preset time window, define the data interval representation of each type of sensor to determine the upper and lower limits of the data within the corresponding time interval to obtain a fusion interval;

[0089] S03. For any two fused intervals, calculate the interval correlation, and also calculate the average correlation of each fused interval with other intervals;

[0090] S04. Map each fusion interval to a non-probabilistic mass distribution of a safety state to complete non-probabilistic modeling;

[0091] S05. Calculate a weighting coefficient based on the average correlation of each fusion interval and weight the quality distribution;

[0092] S06. According to the DS evidence theory, all weighted mass assignments are sequentially integrated to eliminate conflicts and obtain the final composite mass assignment function;

[0093] S07. Calculate the confidence and confidence level based on the final composite quality allocation function and non-probabilistic modeling, and then calculate the comprehensive reliability index;

[0094] S08. Determine the working safety status of the crane based on the comprehensive reliability index.

[0095] In the present solution S01, the monitoring data includes one or more of stress, vibration and temperature, which are collected and fed back by different types of sensors.

[0096] In order to avoid false alarms caused by data collection by a single number of sensors of the same type, in this solution, there can be multiple sensors of each type. In terms of monitoring data, as a possible implementation method, further, in this solution S02, when the same monitoring data is collected by multiple sensors of the same type, the data collected and fed back by the sensors of the same type within the same preset time window are weightedly fused according to preset weights to obtain a fusion interval.

[0097] Among them, the data interval of each type of sensor is expressed as is the sensor number, are the lower limit and upper limit of the data collected by sensor j within the preset time window.

[0098] Specifically, when the same monitoring data is collected by a single sensor, the fusion interval i r =Ij .

[0099] When the same monitoring data is collected by multiple same type sensors, the fusion interval I r is expressed as follows:

[0100]

[0101] wherein j is the sensor number, are respectively the lower limit value and the upper limit value of the data collected by sensor j within the preset time window; w j is the weight value of the data collected by sensor j, and the cumulative result of the weight values corresponding to the multiple same type sensors is 1, that is, N is the number of the multiple same type sensors.

[0102] In this scheme, since the data collected by the multiple source same type sensors has deviation or conflict, directly superimposing it is easy to introduce error, therefore, in this scheme, the fluctuation range of each sensor within the time window is expressed by using interval, and then the intervals are fused by using the preset weight, so that the obtained data interval can represent the reliable interval of the physical quantity.

[0103] In terms of excluding abnormal data, the collected monitoring data is preprocessed in this scheme, as a relatively optimal implementation selection, preferably, in the scheme S01, the way of preprocessing the monitoring data includes filtering and smoothing the monitoring data by using the SG filtering algorithm, and normalizing the data after filtering and smoothing.

[0104] wherein, for the data record aspect, the monitoring data collected and fed back by the sensor is sequentially recorded as X={x i |i=1, 2...m}, x i is the monitoring data collected and fed back by the sensor at the i-th moment, and m is the number of data samples; the data sequence after filtering and smoothing is expressed as:

[0105]

[0106] wherein M≤i-1≤m-M, the filtering window width is 2M+1, c k is the SG filtering coefficient, and satisfies∑c k =1, is the smoothed value of the monitoring data at the i-th moment after filtering; the data record obtained after filtering and smoothing is

[0107] In addition, the data after filtering and smoothing is also normalized in this scheme, which includes:

[0108] the data The minimum-maximum normalization processing is performed in the same time window, and the formula definition is as follows:

[0109]

[0110] wherein, is the monitoring data value of the i th moment after filtering smoothing and normalization processing, is the monitoring data smoothing value of the i th moment after filtering, are the minimum value and the maximum value in the data sequence after smoothing processing respectively.

[0111] After filtering processing, the noise interference in the original sensor signal can be eliminated to a certain extent, and the data of different dimensions are unified to the same scale to ensure the reliability and comparability of subsequent fusion. In order to make the data sequence obtained by the present scheme better for subsequent analysis reuse, and to improve the reliability of the obtained results.

[0112] Based on the need to quantify the consistency or conflict degree between each fusion interval, in order to subsequently dynamically adjust the importance of each evidence, the present scheme calculates the interval overlap rate as the correlation degree, and then obtains the average correlation degree of each evidence (interval). As a relatively optimal implementation selection in data fusion, preferably, the present scheme S03 comprises:

[0113] Any two fusion intervals in the fusion interval are set as The interval correlation degree ε ab is calculated, and the definition is as follows:

[0114]

[0115] wherein, ε ab ∈ [0, 1], the greater the value of ε ab , the more the overlap of the fusion intervals I a and I b , and the stronger the consistency.

[0116] On the basis of the above, the average correlation degree of each fusion interval with other intervals is also calculated , and the definition is as follows:

[0117]

[0118] wherein, K is the total number of monitoring data types to be analyzed after data fusion; is the consistency average value of the k th fusion interval with other fusion intervals.

[0119] ​To convert the interval form of uncertainty information into the "mass distribution" (belief mass) required by the D-S evidence theory, the present scheme fuses the interval mapping into a mass triplet supporting "safe" (H), supporting "failure" and uncertainty (Θ).

[0120] Specifically, in terms of non-probabilistic reliability modeling, as a preferred implementation option, the present scheme S04 comprises:

[0121] Each fusion interval is set as mapping it to the non-probabilistic mass distribution m k of the safe state H to complete the non-probabilistic modeling, which is defined as follows:

[0122]

[0123] where S max is the preset upper limit of safety, F max is the preset failure threshold, which are the lower and upper limits of the reliability mapping of the fusion interval, respectively; m k (H) is the mass of the kth data evidence supporting "safe", m k (Θ) is the mass representing uncertainty.

[0124] Since different data evidences should not have fixed credibility due to different sources or consistency, it is necessary to dynamically adjust to weaken the influence of high-conflict data. As a preferred implementation option, the present scheme S05 comprises:

[0125] According to the average correlation of each fusion interval to calculate the weighting coefficient α k , which is defined as follows:

[0126]

[0127] where α k is the weight factor of the kth data evidence, i.e., the weight factor of the monitoring data collected by the kth sensor after fusion, which is used to weight the mass distribution m k , α k ∈[0, 1], and satisfies K is the total number of monitoring data types to be analyzed after data fusion; is the average consistency of the kth fusion interval with other fusion intervals, and j is an index variable representing the traversal of K data evidences for calculating the normalization denominator of the weight factor, is the normalization of the inconsistency degree of all data evidences, so that

[0128] In the present solution, the quality distribution m k is weighted, which is defined as follows: k

[0129] m′ k (A) = a k m k (A)

[0130] wherein, m′ k (A) is the weighted quality distribution, a k is the weight factor of the kth data evidence, and m k (A) is the quality distribution of the kth data evidence.

[0131] As a preferred implementation option, preferably, the present solution S06 comprises:

[0132] The conflict factor of any two data evidences is defined as follows:

[0133]

[0134] wherein, K ij is the conflict factor, m′ i (B) and m′ j (C) are the weighted quality distributions of the ith and jth data evidences, respectively.

[0135] In order to effectively eliminate conflicts while retaining the information of each evidence, so that the final fusion result is both comprehensive and robust, the present solution adopts the weighted Dempster-Shafer rule, which first discounts according to the weight, then merges the evidences two by two, and automatically handles the conflicts.

[0136] Specifically, in the present solution, all weighted quality distributions are sequentially fused according to the D-S evidence theory to eliminate conflicts, wherein the function for fusing the quality distributions of two data evidences is defined as follows:

[0137]

[0138] wherein, m ij (A) is the quality distribution after re-fusing the weighted quality distributions of the ith and jth data evidences.

[0139] In the present solution, the fusion of the quality distributions of data evidences is extended to K data evidences, which are sequentially merged two by two, and finally the final synthesis quality distribution function m is obtained, which is defined as follows:

[0140] ​

[0141] wherein, m' K is the weighted quality assignment of the Kth data evidence.

[0142] In order to extract the measurable "safety confidence" and "plausibility" from the synthesized quality assignment, and give a single reliability index, the scheme calculates Belief, Plausibility, and then synthesizes the comprehensive reliability R with the intermediate weight (usually 0.5).

[0143] Specifically, as a preferred implementation option, the scheme S07 includes:

[0144] According to the final synthesized quality assignment function m and the non-probabilistic modeling calculation confidence Bel(H) and plausibility PI(H), which are defined as follows:

[0145]

[0146] According to the calculated confidence and plausibility, further calculate the comprehensive reliability index, which is defined as follows:

[0147]

[0148] wherein, Bel(H) ∈ [0, 1], which represents the safety confidence in the most conservative state, PI(H) ∈ [0, 1], which represents the safety confidence in the most optimistic state, and the comprehensive reliability index R ∈ [0, 1].

[0149] As a preferred implementation option, the scheme S08 includes:

[0150] Define the correspondence between the comprehensive reliability index and the working safety state of the crane, which includes the following:

[0151]

[0152] According to the comprehensive reliability index, the working safety state of the crane is judged.

[0153] In order to facilitate further illustration of the scheme, the scheme is explained in combination with the following data examples, and the data collection belongs to the conventional means. The present example scheme is directly explained from S02. Assuming that after the data collection and preprocessing of S01, the following three types of data fusion intervals are obtained through the data fusion of S02:

[0154]

[0155] wherein, the preset safety upper limit value S max is 0.2, and the preset failure threshold Fmax is 1.0.

[0156] The consistency between each fusion interval is measured, i.e. interval correlation degree ε ab The calculation is as follows:

[0157]

[0158] 1. Stress-vibration

[0159]

[0160] 2. Stress-temperature

[0161]

[0162] 3. Vibration-temperature

[0163]

[0164] On the basis of the above, the average correlation degree of each fusion interval with other intervals is also calculated The definition is as follows:

[0165]

[0166] Wherein, K=3, the specific calculation is as follows:

[0167]

[0168] According to the average correlation degree of each fusion interval The weighted coefficient a is calculated k The definition is as follows:

[0169]

[0170] Specifically:

[0171]

[0172] a 应力 = 0.2834, a 温度 = 0.3248, a 振动 = 0.3918

[0173] Each fusion interval is mapped to a non-probabilistic mass distribution m of the safety state H k To complete the non-probabilistic modeling, the definition is as follows:

[0174]

[0175] Wherein, for stress, it includes the following:

[0176] For vibration, it includes as follows:

[0177]

[0178] m 振动 (H) = 0.375

[0179]

[0180] For temperature, it includes as follows:

[0181]

[0182] m 温度 (H) = 0.25

[0183]

[0184] By weighting coefficient a k The above calculation results are weighted, which includes:

[0185]

[0186] The data evidence of stress and vibration object is fused first, and the conflict factor K 12 = K 应力-振动 , which is calculated as follows:

[0187]

[0188] The synthetic mass distribution is carried out again, which is calculated as follows:

[0189]

[0190] The data quality distribution m 12 of stress and vibration after fusion is fused with temperature data evidence, and the conflict factor K (12)，3 = K (应力-温度)-温度 , which is calculated as follows:

[0191]

[0192] The synthetic mass distribution is carried out again, which is calculated as follows:

[0193]

[0194] According to the final synthetic mass distribution function m and the non-probabilistic modeling calculation confidence Bel(H) and plausibility PI(H), which are defined as follows:

[0195]

[0196] According to the calculated reliability and the confidence, a comprehensive reliability index is further calculated, which is defined as follows:

[0197]

[0198] According to the calculation as above, 0.5≤R<0.8, thus, the evaluation result is that the monitoring is enhanced, and the reliability monitoring of the crane operation can be improved by increasing the monitoring frequency or the project.

[0199] In combination with Figure 2 based on the above, the scheme further proposes a crane non-probabilistic reliability collaborative evaluation system based on multi-source data fusion, which comprises:

[0200] The data collection module collects the monitoring data fed back by the sensor group on the crane when the crane is in a working state, and then pre-processes the monitoring data;

[0201] The data fusion module is used to define the data interval of each type of sensor within a preset time window, to determine the upper and lower limits of the data within the corresponding time interval, and to obtain the fusion interval;

[0202] The data processing unit is used to calculate the interval correlation of any two fusion intervals, and also to calculate the average correlation of each fusion interval with other intervals;

[0203] The data modeling module is used to map each fusion interval to a non-probabilistic quality distribution of a safety state, to complete non-probabilistic modeling;

[0204] The data post-processing module calculates a weighting coefficient according to the average correlation of each fusion interval, and weights the quality distribution;

[0205] The data synthesis module is used to sequentially fuse all the weighted quality distributions according to the D-S evidence theory, to eliminate conflicts, and to obtain a final synthesized quality distribution function;

[0206] The reliability calculation module is used to calculate the reliability and the confidence according to the final synthesized quality distribution function and the non-probabilistic modeling, and to further calculate a comprehensive reliability index;

[0207] The safety evaluation module is used to judge the safety condition of the crane operation according to the comprehensive reliability index.

[0208] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0209] If the integrated unit is implemented in the form of a software function unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such understanding, the technical solutions of the present application or the entire or part of the technical solutions that essentially contribute to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to execute all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0210] The above only describes some embodiments of the present application, and does not limit the protection scope of the present application. Any equivalent device or equivalent process transformation, or direct or indirect application in other related technical fields based on the content of the present application specification and drawings are also included in the patent protection scope of the present application.

Claims

1. A collaborative evaluation method for crane non-probabilistic reliability based on multi-source data fusion, characterized by: It includes: S01. When the crane is in operation, collect monitoring data fed back by the sensor group on the crane and then pre-process it; S02. Within a preset time window, define the data interval representation of each type of sensor to determine the upper and lower limits of the data within the corresponding time interval to obtain a fusion interval; S03. For any two fused intervals, calculate the interval correlation, and also calculate the average correlation of each fused interval with other intervals; S04. Map each fusion interval to a non-probabilistic mass distribution of a safety state to complete non-probabilistic modeling; S05. Calculate a weighting coefficient based on the average correlation of each fusion interval and weight the quality distribution; S06. According to the DS evidence theory, all weighted mass assignments are sequentially integrated to eliminate conflicts and obtain the final composite mass assignment function; S07. Calculate the confidence and confidence level based on the final composite quality allocation function and non-probabilistic modeling, and then calculate the comprehensive reliability index; S08. Determine the working safety status of the crane based on the comprehensive reliability index.

2. The crane non-probabilistic reliability collaborative evaluation method based on multi-source data fusion according to claim 1 is characterized in that: In S01, the monitoring data includes one or more of stress, vibration, and temperature, which are collected and fed back by different types of sensors; In S02, when the same monitoring data is collected by multiple sensors of the same type, the data collected and fed back by the sensors of the same type within the same preset time window are weightedly fused according to preset weights to obtain a fusion interval; Among them, the data interval of each type of sensor is expressed as j is the sensor number, are the lower limit and upper limit of the data collected by sensor j within the preset time window; When the same monitoring data is collected by a single sensor, the fusion interval I r =I j ; When the same monitoring data is collected by multiple sensors of the same type, the fusion interval I r It is expressed as follows: Where j is the sensor number, are the lower limit and upper limit of the data collected by sensor j within the preset time window; w j is the weight value of the data collected by sensor j. The cumulative weight values ​​corresponding to multiple similar sensors are 1, that is, N is the number of sensors of the same type.

3. The crane non-probabilistic reliability collaborative evaluation method based on multi-source data fusion according to claim 2 is characterized in that: In S01, the monitoring data is pre-processed by using an SG filtering algorithm to filter and smooth the monitoring data, and normalizing the filtered and smoothed data. Among them, the monitoring data collected and fed back by the sensor are recorded in sequence as X={x i |i=1,2......m},x i is the monitoring data collected and fed back by the sensor at the i-th moment, and m is the number of data samples; The data sequence after filtering and smoothing is expressed as: Among them, M≤i-1≤mM, the filter window width is 2M+1, c k is the SG filter coefficient, satisfying ∑c k =1, is the smoothed value of the monitoring data at the i-th moment after filtering; after filtering and smoothing, the data record obtained is Normalization of the filtered and smoothed data includes: Data Perform minimum-maximum normalization within the same time window, and the formula is defined as follows: in, It is the monitoring data value at the i-th moment after filtering, smoothing and normalization. is the smoothed value of the monitoring data at the i-th moment after filtering, They are the data series after smoothing The minimum and maximum values ​​in .

4. The crane non-probabilistic reliability collaborative evaluation method based on multi-source data fusion according to claim 3 is characterized in that: S03 includes: Set any two fusion intervals in the fusion interval to Calculate its interval correlation ε ab , defined as follows: Among them, ε ab ∈[0, 1], ε ab The larger the value, the closer the fusion interval I a , I b The more overlap there is, the stronger the consistency between the two; The average correlation between each fusion interval and other intervals is also calculated It is defined as follows: Where K is the total number of monitoring data types to be analyzed after data fusion; is the average consistency between the k-th fusion interval and other fusion intervals.

5. The crane non-probabilistic reliability collaborative evaluation method based on multi-source data fusion according to claim 4 is characterized in that: S04 includes: Set each fusion interval to Map it into a non-probabilistic mass assignment m to the safe state H k , to complete the non-probabilistic modeling, which is defined as follows: Among them, S max is the preset safety upper limit, F max is the preset fault threshold, They are the lower and upper limits of the fusion interval mapped to the reliability; m k (H) is the quality of the kth category of data evidence supporting "safety", is the quality of the k-th type of data evidence supporting "failure", m k (θ) is the mass representing uncertainty.

6. The crane non-probabilistic reliability collaborative evaluation method based on multi-source data fusion according to claim 1 is characterized in that: S05 includes: According to the average correlation of each fusion interval Calculate the weighting coefficient α k , which is defined as follows: Among them, α k is the weight factor of the kth piece of data evidence, that is, the weight factor of the monitoring data collected by the kth sensor after fusion, which is used to assign quality m k Weighted, α k ∈[0,1], and satisfies K is the total number of monitoring data types to be analyzed after data fusion; is the average consistency between the k-th fusion interval and other fusion intervals, j is an index variable, which represents the traversal of K data evidences for calculating the normalized denominator of the weight factor, In order to normalize the inconsistency of all data evidence, By weighting coefficient α k Assign mass m k The weighting is defined as follows: m′ k (A)=a k m k (A) in, m′ k (A) is the weighted mass distribution, α k is the weight factor of the kth piece of data evidence, m k (A) is the quality distribution of the k-th piece of data evidence.

7. The crane non-probabilistic reliability collaborative evaluation method based on multi-source data fusion according to claim 6 is characterized in that: S06 includes: The conflict factor of any two pieces of data evidence is defined as follows: in, K ij is the conflict factor, m′ i (B), m′ j (C) The weighted quality distribution of the i-th and j-th data evidence respectively; According to the DS evidence theory, all weighted quality assignments are fused in sequence to eliminate conflicts. The function for fusing the quality assignments of two pieces of data evidence is defined as follows: in, m ij (A) is the quality distribution after re-integrating the weighted quality distribution of the i-th and j-th data evidence; The quality distribution fusion of data evidence is extended to K pieces of data evidence, and each pair is merged in turn to obtain the final composite quality distribution function m, which is defined as follows: Among them, m′ K is the weighted quality distribution of the K-th data evidence.

8. The crane non-probabilistic reliability collaborative evaluation method based on multi-source data fusion according to claim 7 is characterized in that: S07 includes: The confidence Bel(H) and confidence PI(H) are calculated based on the final composite quality distribution function m and non-probabilistic modeling, which are defined as follows: Based on the calculated confidence and confidence level, the comprehensive reliability index is calculated, which is defined as follows: Among them, Bel(H)∈[0,1] represents the safety confidence under the most conservative state, PI(H)∈[0,1] represents the safety confidence under the most optimistic state, and the comprehensive reliability index R∈[0,1].

9. The crane non-probabilistic reliability collaborative evaluation method based on multi-source data fusion according to claim 8, characterized in that: S08 includes: The corresponding relationship between the comprehensive reliability index and the crane working safety status is defined as follows: The safety status of the crane operation is judged based on the comprehensive reliability preparation.

10. A crane non-probabilistic reliability collaborative evaluation system based on multi-source data fusion, characterized in that: It includes: The data collection module collects the monitoring data fed back by the sensor group on the crane when the crane is in operation, and then pre-processes it; The data fusion module is used to define the data interval representation of each type of sensor within a preset time window to determine the upper and lower limits of the data within the corresponding time interval and obtain the fusion interval; A data processing unit is used to calculate the interval correlation between any two fused intervals and also calculate the average correlation between each fused interval and other intervals; A data modeling module is used to map each fusion interval into a non-probabilistic mass distribution of a safety state to complete non-probabilistic modeling; The data post-processing module calculates the weighting coefficient based on the average correlation of each fusion interval and weights the quality distribution; The data synthesis module is used to sequentially fuse all weighted mass assignments according to the DS evidence theory to eliminate conflicts and obtain the final composite mass assignment function; A reliability calculation module is used to calculate the reliability and confidence level based on the final composite quality distribution function and non-probabilistic modeling, and then calculate the comprehensive reliability index; The safety assessment module is used to judge the working safety status of the crane based on the comprehensive reliability index.

Images